Flight heights obtained from GPS versus altimeters influence estimates of collision risk with offshore wind turbines in Lesser Black-backed Gulls Larus fuscus

Abstract The risk posed by offshore wind farms to seabirds through collisions with turbine blades is greatly influenced by species-specific flight behaviour. Bird-borne telemetry devices may provide improved measurement of aspects of bird behaviour, notably individual and behaviour specific flight heights. However, use of data from devices that use the GPS or barometric altimeters in the gathering of flight height data is nevertheless constrained by a current lack of understanding of the error and calibration of these methods. Uncertainty remains regarding the degree to which errors associated with these methods can affect recorded flight heights, which may in turn have a significant influence on estimates of collision risk produced by Collision Risk Models (CRMs), which incorporate flight height distribution as an input. Using GPS/barometric altimeter tagged Lesser Black-backed Gulls Larus fuscus from two breeding colonies in the UK, we examine comparative flight heights produced by these devices, and their associated errors. We present a novel method of calibrating barometric altimeters using behaviour characterised from GPS data and open-source modelled atmospheric pressure. We examine the magnitude of difference between offshore flight heights produced from GPS and altimeters, comparing these measurements across sampling schedules, colonies, and years. We found flight heights produced from altimeter data to be significantly, although not consistently, higher than those produced from GPS data. This relationship was sustained across differing sampling schedules of five minutes and of 10 s, and between study colonies. We found the magnitude of difference between GPS and altimeter derived flight heights to also vary between individuals, potentially related to the robustness of calibration factors used. Collision estimates for theoretical wind farms were consequently significantly higher when using flight height distributions generated from barometric altimeters. Improving confidence in telemetry-obtained flight height distributions, which may then be applied to CRMs, requires sources of errors in these measurements to be identified. Our study improves knowledge of the calibration processes for flight height measurements based on telemetry data, with the aim of increasing confidence in their use in future assessments of collision risk and reducing the uncertainty over predicted mortality associated with wind farms.